intelligent active force control of a human-like...

INTELLIGENT ACTIVE FORCE CONTROL OF A HUMAN-LIKE ARM

ACTUATED BY PNEUMATIC ARTIFICIAL MUSCLES

HOSSEIN JAHANABADI

A thesis submitted in fulfilment of the

requirements for the award of the degree of

Master of Engineering (Mechanical)

Faculty of Mechanical Engineering

Universiti Teknologi Malaysia

APRIL 2010

iii

DEDICATION

I would like to dedicate this thesis to:

My beloved parents, Mohammad Jahanabadi and Sedeigheh Agheb, to my

brothers Ali and Reza, to my sister Atefeh for their tremendous support,

encouragement and patience throughout my studies and indeed throughout my life.

iv

ACKNOWLEDGEMENTS

I would like to express my deepest gratitude and thanks to my reasearch

supervisor, Professor Dr. Musa Mailah who has provided advice, expertise, excellent

guidance and support throughout the project. I have learned so much over this

research and I greatly appreciate the opportunity to work on such an exciting project,

and to have gained the experience that will help me immensely in my future

endeavors in all aspects of life. I would also like to thank my co-supervisor Dr. Mohd

Zarhamdy Md Zain for his invaluable support, guidance and suggestions throughout

my study.

I would like to thank Mr. Muhamad Akmal Bahrain A Rahim for his precious

help in developing the experimental rig.

Special thanks to the Universiti Teknologi Malaysia for supporting the study

by providing excellent research rersources and facilities that include the laboratories

and library.

Last but not least, I would also like to thank the Ministry of Science,

Technology and Innovation (MOSTI) of Malaysia for giving the financial support for

my research through the E-Sciencefund grant (Project No.: 03-01-06-SF0408).

v

ABSTRACT

Robotic system driven by fluidic muscles is time-varying and exhibits high

degree of nonlinearity due to the internal system behaviours and thus controlling it

constitutes a major problem. This poses a great challenge to researchers to come up

with suitable technique/s to control such system effectively. The research presents a

simulation and experimental study of a force control method applied to a two-link

planar ‘human-like’ robot arm that is actuated by fluidic muscles. Active force

control (AFC) based scheme was particularly implemented to the system

incorporating two types of intelligent techniques, namely, fuzzy logic (FL) and

iterative learning (IL) to effectively and robustly control the arm driven by a

pneumatic artificial muscle (PAM) system and subject to a number of operating and

loading conditions. The PAM is actuated by two groups of fluidic muscles in

bicep/tricep configuration. The simulation and the experimental study verify that

proposed system performs excellently even in the presence of uncertainties,

hysteresis behaviour of the actuator and inherent nonlinearity. Joint trajectory

planning was applied to ensure that the arm tracks the given input commands

accurately considering a number of different frequency settings. The simulated

system was complemented and validated through an experimental study carried out

on a developed rig via a convenient hardware-in-the-loop simulation (HILS)

technique using suitable hardware and software interface. The obtained results both

through simulation and experimental investigation clearly imply the viability of the

proposed AFC-based system in controlling the PAM actuated robotic arm and they

also demonstrate the system superiority over the PID controller alone counterpart.

vi

ABSTRAK

Sistem robotik dipacu oleh otot bendalir yang berubah dengan masa dan

menunjukkan darjah ketidaklelurusan yang tinggi disebabkan oleh kelakuan sistem

dalaman; maka untuk mengawal sistem sedemikian merupakan satu masalah besar.

Ini memberikan satu cabaran kepada para penyelidik untuk menghasilkan satu

kaedah kawalan yang berkesan untuk menangani masalah tersebut. Penyelidikan

yang dijalankan memaparkan suatu kajian simulasi dan eksperimen berkaitan dengan

kaedah kawalan daya terhadap suatu lengan robot planar dua penghubung

menyerupai lengan manusia yang dipacu oleh otot bendalir. Skema berlandaskan

kawalan daya aktif (AFC) digunakan terhadap sistem dengan memuatkan dua jenis

teknik pintar, iaitu logik kabur (FL) dan pembelajaran berlelaran (IL) bagi tujuan

mengawal dengan lasak dan berkesan suatu lengan yang dipacu oleh satu sistem otot

tiruan pneumatik (PAM) yang ditindakkan oleh beberapa keadaan bebanan dan

pengoperasian. PAM digerakkan oleh dua kumpulan otot bendalir dalam konfigurasi

bicep/tricep. Kajian simulasi dan eksperimen mengesahkan bahawa sistem yang

dicadangkan memberikan prestasi yang amat baik walaupun beroperasi dalam

keadaan wujudnya ketidaktentuan, kelakuan histerisis penggerak dan

ketidaklelurusan dalaman. Perancangan trajektori sendi digunakan bagi memastikan

lengan menjejaki dengan tepat arahan masukan yang dikenakan dengan mengambil

kira beberapa nilai frekuensi yang berlainan. Sistem yang dikaji menerusi simulasi

dapat ditentusahkan menerusi kajian eksperimen yang dijalankan terhadap suatu rig

yang dihasilkan menggunakan kaedah simulasi perkakasan-di-dalam-gelung (HILS)

melibatkan penggunaan pengantaramuka perisian dan perkaksan yang bersesuaian.

Hasil keputusan kajian yang dijalankan baik menerusi simulasi mahupun eksperimen

menunjukkan bahawa sistem berlandaskan AFC yang dicadangkan berupaya

viii

TABLE OF CONTENTS

CHAPTER TITLE PAGE

DECLARATION ii

DEDICATION iii

ACKNOWLEDGEMENTS iv

ABSTRACT v

ABSTRAK vi

TABLE OF CONTENTS viii

LIST OF TABLES xi

LIST OF FIGURES xii

LIST OF ABBREVIATIONS xv

LIST OF SYMBOLS xvi

LIST OF APPENDICES xviii

1 INTRODUCTION

1.1 General Introduction........................................................................1

1.2 Research Background and Contributions..........................................2

1.3 The Problem Statements ..................................................................4

1.4 Research Objective ..........................................................................5

1.5 Research Scope................................................................................6

1.6 Research Methodology ....................................................................6

1.6.1 Modelling and Simulation ....................................................7

1.6.2 Controller design..................................................................7

1.6.3 Design and Development of the Experimental Rig................8

1.6.4 Experimentation ...................................................................9

ix

1.6.5 Performance Evaluation and Analysis ................................10

1.7 Organization Thesis .......................................................................11

2 LITERATURE REVIEW AND THEORETICAL FRAMEWORK.....13

2.1 Introduction ...................................................................................13

2.2 Review on Related Research..........................................................13

2.3 Pneumatic Artificial Muscle (PAM)...............................................17

2.3.1 PAM Basic Operation ........................................................19

2.3.2 Comparison of Actuators....................................................20

2.3.3 Comparison of PAM to Biological Muscle .........................22

2.3.4 PAM Modelling .................................................................23

2.4 Mathematical Model of the Robotic Arm.......................................26

2.4.1 The Kinematics Model of the Robotic Arm ........................27

2.4.2 The Dynamic Model of the Robotic Arm............................28

2.4.3 Actuator Made of Two Antagonistic Fluidic Muscles .........29

2.5 Active Force Control .....................................................................33

2.5.1 Estimated Inertia Matrix.....................................................35

2.5.2 Estimation of Inertia Matrix using Intelligent Schemes.......36

2.6 Iterative Learning...........................................................................36

2.6.1 Active Force Control and Iterative Learning Scheme

(AFCAIL) ..........................................................................38

2.7 Fuzzy Logic...................................................................................41

2.7.1 Fuzzy Sets and Membership Functions...............................42

2.7.2 Linguistic Variables ...........................................................43

2.7.3 If-Then Rules .....................................................................44

2.7.4 Fuzzy Inference Systems....................................................46

2.7.5 Active Force Control and Fuzzy Logic (AFCAFL) .............51

2.8 Conclusion.....................................................................................51

3 SIMULATION OF A TWO-LINK PAM ACTUATED ROBOT

SYSTEM..................................................................................................53

3.1 Introduction ...................................................................................53

3.2 The Proposed Active Force Control and Iterative Learning Scheme54

3.2.1 Simulation Parameters........................................................56

x

3.2.2 Initial Parameters ...............................................................57

3.2.3 Sampling Time...................................................................58

3.2.4 Stopping Criteria ................................................................58

3.2.5 Applied Disturbance...........................................................58

3.3 The Proposed Active Force Control and Fuzzy Logic Scheme .......59

3.3.1 Simulation Parameters........................................................59

3.3.2 Definition of Linguistic Variables .....................................60

3.3.3 Determination of Fuzzy Sets 61

3.3.4 Design of Fuzzy Rules…………………………………… 63

3.3.5 Deffuzification…………………………………………… 66

3.4 Results and Discussion ..................................................................66

3.5 Conclusion.....................................................................................73

4 EXPERIMENTAL ROBOT ARM MANIPULATOR...........................75

4.2 Introduction ...................................................................................75

4.2 Mechanical-Based Design..............................................................76

4.3 Computer-Based Control ...............................................................79

4.4 Sensors and Actuators....................................................................82

4.4.1 Accelerometer ....................................................................82

4.4.2 Encoder..............................................................................83

4.4.3 Proportional Pressure Regulator .........................................84

4.4.4 Fluidic Muscle ...................................................................84

4.5 Experimental Results and Discussion.............................................85

4.6 Conclusion.....................................................................................95

5. CONCLUSION AND FUTURE WORKS ……………………………….96

5.1 Conclusion…………………………………………………………..96

5.2 Recommendations for Future Works………………………………..97

REFERENCES……………………………………………………………… 99

APPENDICES 106

xi

LIST OF TABLES

TABLE NO. TITLE PAGE

2.1 A comparison between different types of actuators 21

3.1 Acceleration 60

3.2 Trackerror………………………………………………………………... 60

3.3 Estimated inertia element IN1……………………………………………… 61

3.4 Estimated inertia element IN2……………………………………………… 61

3.5 Fuzzy rules…………………………………………………………………..64

4.1 Specifications of the rig……………………………………………………..79

4.2 Specifications of PCI-1710HG…………………………………………. 81

xii

LIST OF FIGURES

FIGURE NO. TITLE PAGE

1.1 Implementation of the research project 10

2.1 Basic operation of a pneumatic artificial muscle 19

2.2 Geometry of Festo Actuator. The braided fiber turns n times 24

2.3 Application of the virtual works theorem to the PAM 25

2.4 A representation of a two link arm 27

2.5 Skeleto-Muscle system of human elbow joint 30

2.6 Working Principle of antagonistic fluidic muscle actuator 30

2.7 Working Principle of antagonistic fluidic muscle actuator for two link arm.32

2.8 A general form of AFC scheme 33

2.9 A schematic diagram of proposed PID-AFC 34

2.10 PD type scheme 38

2.11 The interlinking of the proposed control scheme 39

2.12 A proposed AFCAIL scheme using a PD type algorithm ......................... 40

2.13 A flowchart showing the PD type learning algorithm applied to robot arm 41

2.14 Interpreting if-then rules 46

2.15 Graphical representation of the introductory example 47

2.16 Fuzzifying inputs 48

2.17 Application of fuzzy operator 49

2.18 Implication (minimum) method 49

2.19 Aggregate of all the individual outputs 50

2.20 Defuzzificarion 50

2.21 The proposed AFCAFL scheme 51

3.1 The Simulink diagram of AFCAIL 54

3.2 The Simulink diagram of Robot’s Dynamic model 55

3.3 The Simulink diagram of AFC 55

xiii

3.4 A window showing the equation representing the fluidic muscle 55

3.5 The Simulink diagram of AFCAFL 59

3.6 Angular acceleration as the input function 62

3.7 Trackerror as the input function 62

3.8 Estimated inertia matrix (IN1) as the output function 63

3.9 Estimated inertia matrix (IN2) as the output function 63

3.10 Rule editor 65

3.11 Surface viewer 65

3.12 Applied disturbance on each joint 68

3.13 Pulse input of link1 68

3.14 Pulse input of link2 68

3.15 Sine input of link1 69

3.16 Sine input of link2 69

3.17 PID performance of link 1 with pulse input 69

3.18 PID performance of link 2 with pulse input 70

3.19 PID performance of link 1 with sine input 70

3.20 PID performance of link 2 with sine input 70

3.21 AFCAIL performance of link 1 with pulse input 71

3.22 AFCAIL performance of link 2 with pulse input 71

3.23 AFCAIL performance of link 1 with sine input 71

3.24 AFCAIL performance of link 2 with sine input 72

3.25 AFCAFL performance of link 1 with pulse input 72

3.26 AFCAFL performance of link 2 with pulse input 72

3.27 AFCAFL performance of link 1 with sine input 73

3.28 AFCAFL performance of link 2 with sine input 73

4.1 Experimental rig made to test the fluidic muscle behaviour 77

4.2 The isometric view of the experimental rig using fluidic muscle as actuator77

4.3 experimental rig of two link planar robot arm 78

4.4 close-up view of the joint 2 78

4.5 The PCI-1710HG advantech 80

4.6 The mounting of the PCI-1710HG card in the PC 80

4.7 Schematic diagram of the experimental set-up 81

4.8 The ADXL330 Accelerometer 83

4.9 The Omron E6B2-C rotary encoder 83

xiv

4.10 Schematic and photo of the proportional pressure regulator 84

4.11 The fluidic muscles (DMAS models) manufactured by FESTO 85

4.12 Performance of link 1 using PID controller, f=0.05 Hz 86


4.14 Performance of link 1 using AFCAIL controller, f=0.05 Hz 87


4.16 Performance of link 1 using AFCAFL controller, f=0.05 Hz 88






4.22 Performance of link 1 using AFCAFL controller, f = 0.05 Hz 90


4.24 Performance of link 1 using PID controller, f = 0.1 Hz 91


4.26 Performance of link 1 using AFCAIL controller, f = 0.1 Hz 92










xv

LIST OF ABBREVIATIONS

AFC Active force control

AFCAFL Active force control and fuzzy logic

AFCAIL Active force control and iterative learning

DOF Degree of freedom

FL Fuzzy logic

HILS Hardware-in-the-loop simulation

IL Iterative learning

PAM Pneumatic artificial muscle

PID Proportional-Integral-Derivative

xvi

LIST OF SYMBOLS

P Absolute gas pressure

dV Volume change

F Actuator force

dl Actuator length change

� Wweave angle

b Fibre length

l Nominal length of PAM

D Diameter of PAM

r Radius

P Internal pressure of PAM

� Contraction ratio

�1 Joint angle of link1


l1 Length of link 1

l2 Length of link 2

m1 Mass of link1



m1 Mass of link1

m2 Mass of link2

lc1 Distance to the center of mass of the link 1

lc2 Distance to the center of mass of the link 2

� Actuated torque vector

H N × N inertia matrix manipulator

xvii

h Coriolis and centripetal torque vector

G Gravitational torque vector

dτ External disturbance torque vector

G Open-loop gain

IN Estimated inertia matrix

*dτ Estimated disturbance torque

θ�� Actual acceleration

Ki Multiplier constant

e Track error

yk+1 Next step value of the output

yk Current output value

ek Current positional error input

� Proportional learning parameter

� Derivative learning parameter

� Integral learning parameter

INk+1 Next step value of estimated inertia matrix

INk Current estimated inertia matrix

TEk Current track error

xviii

LIST OF APPENDICES

APPENDIX TITLE PAGE A Mechanical Parts of the Experimental Rig 106

B Further Results: Performance of Various Control Schemes with f = 0.2 Hz 111

C Publications 115

CHAPTER 1

INTRODUCTION

1.1 General Introduction

Engineering principles are commonly applied to mechanical or

biomechanical systems to better understand the dynamics of the human arm. One

such system is the robotic arm or manipulator which is normally and geometrically

configured to resemble a human arm. A robust and stable performance of a robot arm

is essential as it handles the capability of the arm to compensate for the disturbance

effect, uncertainties, parametric and non parametric changes, which are prevalent in

the system particularly when the arm is executing tasks involving the interaction of

the robot’s end-effector with the environment and some kind of external

disturbances. The coordinated motion and force control of a robot arm is an

important subject area of research, which can directly contribute to the

accomplishment of such objective. A very desirable and effective robot control

system is one in which the issue of robustness is well accounted for. Many robots

control methods have been proposed such as proportional-integral-derivative (PID)

control [1], computed-torque control [2], intelligent control [3], and active force

control (AFC) [4]. It is a well-known fact that the conventional PID control is the

most widely and practically used scheme in the industrial robots due to its good

stability characteristic, simple controller structure, and reliability [1]. It provides a

2

medium to high performance when it comes to robot’s operation at relatively low

speed with little or no disturbance effects. On the contrary, the performance suffers

severe setbacks when adverse conditions prevail. A number of researches have been

conducted to seriously address the issue and determine ways to counter the weakness

[5, 6]. Thus, a feedback control system should be ideally chosen so that it is able to

overcome most if not all of the drawbacks and at the same time, it is practically

viable and can be implemented in real-world applications. Considering these factors,

the undertaken research project is an attempt to highlight the potentials of an AFC-

based method applied to a two-link planar robot arm actuated by pneumatic artificial

muscle (PAM) actuators.

1.2 Research Background and Contributions

Almost all the robot control methods contained the classical elements (of the

PID control), which contribute to the better overall performance of the system. One

such robot control method, which is of particular interest, is the active force control

strategy first proposed by Hewit in the late seventies [4]. The main feature of this

type of control method is the potential of the AFC concept to dynamical systems

including robot control. By implementing the AFC to the system, the effects due to

any known or unknown disturbances (internal or external), parametric changes and

varied operating conditions can be significantly compensated or eliminated. The

AFC method involves a direct measurement of the acceleration and force quantities

plus the appropriate estimation of the inertia matrix of effect in its control strategy.

Actuators are indispensable for all robots to provide forces, torques, and

mechanical motions to move the joint, limb or body. Actuators are generally electric,

pneumatic, or hydraulic. Today’s mechanical systems have such criteria for actuators

including high power density, high power to weight ratio, rapid response, accurate

and repeatable control, low cost, cleanliness and high efficiency. An important area

of robotics technology is concerned with the development of manipulators that can

3

replace human beings in the execution of specific tasks. This makes such qualities

such as lightweight, high power, and fast accurate response even more important for

actuators. The pneumatic artificial muscle (PAM) actuator possesses many of these

advantages, which therefore considered as an excellent candidate for robotic

applications. However, the inherent nonlinearities, time-varying parameters, and high

sensitivity to payload of the PAM make it a challenge for the accurate force and

position control of manipulators in employing these actuators.

It is the central theme of the undertaken research project to investigate the

implementation and the performance of a robot arm that is actuated and controlled by

PAM through the AFC-based scheme. In the proposed research, an intelligent

component is embedded into the AFC main loop to compute the appropriate inertia

matrix of the arm. Both the theoretical and experimental aspects were highlighted in

the study to validate and illustrate the practical viability of the proposed scheme.

The main research contributions can be summarised as follows:

1. The use of pneumatic artificial muscle (PAM) actuators to drive a two-

link robot manipulator that resembles a human arm.

2. Direct application of the AFC technique to the robot arm through both

simulation and experimental studies. The application of AFC to PAM

actuated system is indeed new area of research.

3. Incorporation of intelligent methods using fuzzy logic (FL) and iterative

learning (IL) algorithms to estimate the inertial parameters of the arms in

the AFC loop for the given system.

4. Development of an experimental rig with Hardware-in-the-Loop-

Simulation (HILS) to validate the proposed control scheme.

4

1.3 The Problem Statements

There are two main issues that need to be addressed and resolved in the

project. The first is related to the AFC scheme while the second is concerned with

the utilisation of the PAM actuator. Both are actually inter-related due to the fact that

the PAM actuator is designed as the main actuating element in the proposed control

schemes.

The AFC strategy is regarded very practical and effective if a number of

criteria can be fulfilled as follows:

1. Accurate measurements of the actuator torques and body angular

accelerations could be obtained.

2. A suitable inverse transfer function of the PAM actuator can be derived

and implemented.

3. A good estimated inertia matrix in the AFC loop can be computed.

The first criterion could be realized easily either through theoretical

investigation through simulation or experimental means using accurate sensors. The

second part is complex as it involves the application of the non-linear PAM and the

required function is derived based on the differential pressure of PAMs, �P. It is

typical that the third criterion is considered the most critical since it involves directly

the ever changing dynamics of a system and hence shall be described in greater detail

here. The estimated inertia matrix of the robotic arm can be derived from the

dynamic model as a function of joint angles which varies as the arm moves.

Mathematically, the computation of this matrix becomes increasingly difficult with

the corresponding increase in the number of degree-of-freedom (DOF) of the robot

arm. It is thus clear that the utilization of the inertia matrix based on the perfect

model of the robot arm with many DOF is quite impractical as far as the AFC

scheme is concerned. For a single DOF case, the solution to the above scheme is

rather straightforward and linear control theory is readily applicable. The proposed

research work assumes a two-link arm (and hence a two DOF system) which is

coupled and inherently non-linear - hence the computation of the estimated inertia

5

matrix is becoming increasingly difficult. It is evident that different system dynamics

produce correspondingly different operating environment related to the acquisition of

the estimated inertial parameters of the arm. Intelligent methods were proposed in the

study as the inertia matrix estimators of the arm driven by the highly non-linear PAM

actuators.

The other aspect is the use of the PAM actuator in the design and

development of the system. Research on the use of PAM actuator for controlling

purpose is relatively new and still in its infancy stage. The main characteristics of the

PAM actuator are the inherent non-linearities, time varying parameters and

sensitivity in the loading of the PAM system that posed great difficulty and challenge

in producing good coordinated motion control of a dynamic system including the

proposed robot arm. This is further exacerbated by the fact that there are not many

published works related to the area of study. On the whole, since the two-link planar

mechanical arm itself is a very nonlinear system plus the use of PAM actuators, the

proposed AFC–based scheme should be designed to be reliable, robust and effective

to tackle the aforementioned problems.

1.4 Research Objective

The main objective of the research is to investigate the performance of the

robotic arm actuated by fluidic muscle actuator through the application of AFC

strategy. To achieve this, the following sub-objectives are considered:

- To carry out theoretical study involving the modelling, simulation and

control of a two-link planar robot arm actuated by PAM.

- To validate the proposed AFC-based scheme through practical

experimentations on a developed rig.

6

1.5 Research Scope

The scope of the undertaken research is as follows:

• A two DOF planar robot arm is investigated considering motion in a horizontal

plane. Hence, gravitational effect can be totally ignored.

• Each joint of the arm is driven by a PAM actuator.

• The arm is assumed to operate at low speed and frequency due to the hysteresis

of the PAM. The only disturbances considered in the study are due to the

inherent non-linearity (including hysteresis) and other time varying parameters

(pressure and contraction ratio) of the PAM actuator.

• Theoretical design and simulation of a robust motion control of the arm is

based on AFC strategy. A class of PID control method is also employed for

benchmarking.

• Two intelligent methods using FL and IL algorithms are utilised to compute the

estimated inertial parameters of the system within the AFC loop.

• Design and development of an experimental PAM actuated two-link planar

system for validation and comparison.

1.6 Research Methodology

The project is divided into five main stages. The stages are modelling and

simulation, controller design, design and development of the rig, experimentation

and performance analysis. Mechatronic system design approach involving the

synergy of mechanical, electrical/electronics and computer (software) control would

be the main feature of the research methodology. The more detailed description of

the research methodology is described in the following sub sections.

7

1.6.1 Modelling and Simulation

As part of the mechatronic design process and prior to simulation, the system

has to be sufficiently modelled to include rigorous manipulation of the governing

mathematical equations representing the system dynamics and kinematics. The

modelling will take into account realistic and valid assumptions related to the main

dynamics of two-DOF robot arm, PAM actuator and disturbances based on the

classic Newtonian mechanics. Other components such as the controllers (PID and

AFC), reference inputs and the intelligent algorithms were mostly modelled based on

typical procedures described in the literature. Simulation study shall include the

evaluation of the system’s performance and robustness against the highly nonlinear

behaviour, hysteresis, and time-varying parameter of the actuators. Comparative

study between the control strategies shall also be done to provide a useful platform in

determining the best control method. The simulation works shall provide a good

basis for designing and developing the experimental rig in later stages. MATLAB

and Simulink software shall be the main design tool used for the simulation study.

Later, the same simulation can be used through a HILS technique that integrates the

hardware (sensors and actuators) and the software via MATLAB, Simulink and Real

Time Windows (RTW) Target facility.

1.6.2 Controller Design

Two main control methods were employed in the study using PID and AFC

methods. Both involve the use of simple control algorithms. In the former, the

controller parameters were obtained using the heuristic method through One-Value-

8

at-A-Time (OVAT) technique which is deemed sufficient. The robot PID control

(with all the relevant models considered, i.e. the dynamics of the robot arm, PAM

actuator, sensor, etc.) can be executed to observe the system performance.

Next the AFC technique is coupled directly to the PID controller. The AFC

parameters were acquired through perfect modelling of the sensors (for the torque

and acceleration measurements) and intelligent inertia matrix estimators via the FL

and IL algorithms. The intelligent algorithms were deliberately selected so that their

practical implementation in real-time is feasibly and readily applied to compute the

required parameters continuously and on-line as the system actually operates. Their

role in the proposed AFC-based scheme is indirect and supplementary in the sense

that they do not directly interfere with the controller main algorithm but rather

provide a method to complete the compensation strategy.

1.6.3 Design and Development of the Experimental Rig

The design and development of the proposed system is in-line with the

Mechatronics approach. In this approach, all the main aspects or disciplines of the

system are simultaneously or concurrently designed and integrated unlike the

traditional approach which is typically more sequential and segregated in nature.

This involves the integration of a number of classical engineering disciplines namely,

the mechanical, electrical/electronics and computer-based control. A complete

integration of the mentioned disciplines is thus very essential to realise a mechatronic

product.

• Mechanical: Mechanical design initially involves the conceptual design of

the physical two-link planar arm. A number of factors and suitable design criteria

should be carefully considered in the design process. The design process will include

conceptual and final design of the robot involving the development of suitable

mechanisms, static and dynamic analysis of the structure of the system, selection of

9

materials and others which are all referred to and should comply with the pre-

determined design criterion. A finalized design will be subsequently chosen with the

detailed production drawings ensued for fabrication purposes. Some of the

mechanical aspects of the system such as the computation of the parameters for the

length of pneumatic artificial muscle, dimensions of the system structure or parts can

be obtained and/or manipulated from the simulation study. The design should also

take into account the ease of the fabrication of the parts to be processed.

• Electrical/Electronics: The selection of the actuator system should be based

on the force analysis of the system to ensure proper actuation of the system is

achieved. Pneumatic artificial muscle-with high power-weight ratio, compliant

nature and whereas their behavior is like muscle can be a suitable actuator to derive a

robotic arm. A good knowledge and skill in electronics assembly is highly desirable

at this stage. Again, the outcome from the simulation works help in determining for

example the size (power, force, torque etc.) of the actuators to be used.

• Computer-based Control: The next stage will involve rigorous computer

interfacing and control involving data acquisition process through the DAQ card and

a PC for software control. Matlab and Simulink serve as the link between the

mechanical and electrical/electronics components.

1.6.4 Experimentation

When the mechatronic system prototype is ready, experimental procedures

will be drafted and testing will be done to validate the effectiveness and the

robustness of the control strategy. Possible modification of the system can be carried

out at this stage to enhance the system’s performance. Experimentation will be

rigorously carried out to test the effectiveness of the proposed system. The tests will

take into account various operating and loading conditions.

10

1.6.5 Performance Evaluation and Analysis

The systems performance will be critically evaluated and analyzed based on

the trajectory tracking capability (accurate position control) of the schemes

considering the effect of inherent non-linearities and disturbances in the system. The

trajectory tracking errors produced by the respective control schemes shall be the

basis for the study of the performance in executing the given tasks. This includes the

results obtained from both the simulated and experimental rig developmental

activities. The research outcomes should provide insights and information which

would be useful for future development, improvement and expansion of the system.

In addition to that, suggestion for further research works will also be outlined.

Figure 1.1 shows the flow chart of the proposed research methodology.

Start

Literature Review

Controller Design

Modelling the PAM and the two link robot arm

Simulation study on two link robot arm

Design and development of experimental rig

Experimental study

Report on the study

Performance evaluation and analysis

Figure 1.1: Implementation of the research project

11

1.7 Organization of Thesis

The thesis is organized into five chapters. In Chapter 2, the fundamental

concepts, underlying theories and reviews of the main topics of research pertaining to

modelling of the kinematics and dynamics of the robotic arm and its control,

pneumatic artificial muscle, PID control, AFC, FL and IL algorithms are described.

The modelling of the two link planar resembled as human-like arm and principles,

characteristics and modelling of pneumatic artificial muscle are described as this

provides the basis for the simulation of the control system. Next, the principles of the

PID and the pure AFC method are discussed with special attention focused on the

method to enhance the strategy using intelligent means, namely IL and FL methods.

An inertia matrix estimation method using fuzzy logic for an AFC scheme is

presented. The fuzzy logic mechanism computes the estimated inertia matrix (IN)

automatically and continuously, while the robot arm is tracking the reference input

that is followed by AFC incorporate with iterative learning (AFCIL). IL is embedded

in the AFC strategy to estimate the estimated inertia matrix (IN) to control the

performance of the robot arm.

Chapter 3 presents the simulation study of proposed controllers, AFCIL and

AFCAFL based on some of the theoretical concepts and principles derived from

previous chapter. Then, the comparative study of the proposed control schemes is

described with a particular emphasis laid on the technique used to compute the

estimated inertia matrix and the trajectory track performance of the proposed control

schemes.

The experimental study is done in Chapter 4 to compliment the theoretical

works. Mechatronics concept is applied here as it involves the rigorous integration of

the mechanical, electronics and computer control disciplines.

12

Finally, Chapter 5 concludes the research project and the directions for the

future works are outlined. Some of the additional materials pertaining to sketch of rig

parts, mechatronics parts catalogue, publications and achievements related to the

research are enclosed in the appendices.

intelligent active force control of a human-like...

Documents